Obtaining the full spatial coverage of daily surface ozone fields is challenging because of the sparsity of the surface monitoring network and the difficulty in direct satellite retrievals of surface ozone. We propose an indirect satellite retrieval framework to utilize the information from satellite-measured column densities of tropospheric NO 2 and CH 2O, which are sensitive to the lower troposphere, to derive surface ozone fields. The method is applicable to upcoming geostationary satellites with high-quality NO2 and CH 2O measurements. To prove the concept, we conduct a simulation experiment using a 3-D chemical transport model for July 2011 over the eastern US. The results show that a second order regression using both NO 2 and CH 2O column densities can be an eff ective predictor for daily maximum 8-h average ozone. Furthermore, this indirect retrieval approach is shown to be complementary to spatial interpolation of surface observations, especially in regions where the surface sites are sparse. Combining column observations of NO 2 and CH 2O with surface site measurements leads to an improved representation of surface ozone over simple kriging, increasing the R 2 value from 0.53 to 0.64 at a surface site distance of 252 km. The improvements are even more significant with larger surface site distances. The simulation experiment suggests that the indirect satellite retrieval technique can potentially be a useful tool to derive the full spatial coverage of daily surface ozone fields if satellite observation uncertainty is moderate. 1. Introduction
Ozone (O3) in the boundary layer is mainly produced from the photochemical reactions of nitrogen dioxide (NO2) and volatile organic compounds (VOC). Harmful to the health of human (Brunekreef and Holgate, 2002) and vegetation (Reich and Amundson, 1985), ozone near the surface has long been thought as an air pollution concern. Information on surface ozone with a full spatial coverage are important for evaluating the ozone exposure to human and vegetation (Avnery et al., 2011; Zou et al., 2009). However, the current ground-based monitoring networks do not provide adequate information to represent the spatial distribution of surface ozone, especially in rural or remote areas. Although there are monitoring networks designed to represent the regional background (e.g., CASTNET), their sites are often too sparse to capture the spatial variability in many regions.
Alternatively, satellite-based observations can provide good spatial coverages. For example, satellite observed aerosol optical depth has been applied to derive daily “ gap-free” surface PM 2.5 field (Lv et al., 2016). Satellite can measure ozone in ultraviolet (UV) and thermal infrared (TIR) bands. However, because of molecular scattering in the
